This paper discusses the importance of high-quality data required for the successful development and implementation of AI-based Industrial Analytics Business solutions.Advanced analytic systems use a variety of data, which include free-text unstructured, free-text structured, categorical, and numerical data.Data-centric artificial intelligence (data-centric AI) is an emerging concept that underscores the importance of developing and continuously enhancing the quality of data to build and maintain efficient AI-based systems.Data-centric AI nicely complements model-centric AI, which focuses on the development and improvement of AI-based systems based on changes and improvements in the AI models keeping the set of data fixed.This paper discusses the importance of developing capabilities to improve the data quality of AI-based systems within the context of data-centric AI.Special emphasis is placed in ensuring the data fed to the AI solution from original data sources meets the required and expected quality.Examples on inherent data quality issues discovered when building and maintaining AI-based Industrial Analytics Business systems are provided and discussed within the concept of data-centric-AI.Moreover, a discussion on how data quality improvements are made within these systems is presented.

Industrial organizations generate massive volumes of data during their routine business and production activities. Such data may be structured (numerical or categorical), or it may be unstructured and textual. Both structured and unstructured data contain a wealth of knowledge that can help organizations improve their operations. Organizations find it easy to automatically extract knowledge from structured data. Unstructured data, however, must be mined and interpreted manually which is cumbersome, error-prone and time consuming. This paper focuses on how to automatically analyze unstructured text data to extract important business value. It proposes a semi-supervised natural language (NL) approach to analyze a corpus of documents associated with accounts receivable disputes at a large corporation. The name semi-supervised derives from the philosophy underlying the methodology, where a set of categories and the keywords associated with these categories are defined in consultation with the domain experts. Subsequently, these categories and their associated keywords are supplied as input to the algorithm, which classifies the disputes automatically into these predefined categories. The performance of the semi-supervised methodology is very comparable to that of the random forest, which is a supervised learning approach. The paper discusses the benefits of the semi-supervised approach over supervised learning; namely, a considerable reduction in the manual effort to analyze, understand and label training data set, without any noticeable degradation in performance.

Anomalies are data points or a cluster of data points that lie away from the neighboring points or clusters and are inconsistent with the overall pattern of the data. Anomaly detection techniques help distinguish the anomalous observations from the regular ones, and thus provide the basis for developing a standard performance guideline for process control. The process of identifying anomalies becomes complicated in the absence of labeled training data as in supervised learning. Moreover, Euclidean distance between two points is less likely able to reflect the intrinsic structural distance imposed by the underlying manifold structure. In this paper, the authors propose a minimum spanning tree (MST)-based anomaly detection method. The merit of the method is that an MST provides a new distance measure, capable of capturing the relative connectedness of data points/clusters in a complicated manifold, and could be a better (dis)similarity metric, than the simple Euclidean distance, to identify anomalies in unsupervised learning settings. The proposed method is compared with 13 popular anomaly detection methods on 20 benchmark data sets, demonstrating a considerable improvement in its ability of identifying anomalies. Furthermore, the MST-based anomaly detection is applied to the data set from a hydropower turbine and demonstrates remarkable detection competence.

This paper focuses on an industrial experience with software product lines of analytics-enabled solutions, specifically the evolution of the software product line architecture for a Subject Matter Expert Workbench toolset which supports analytic plugins for multiple software product lines. As context, the toolset product line was intended for integration of expert knowledge into a family of industrial asset health applications at runtime. The toolset architecture is now being evolved to build and manage plugins for multiple Industrial Analytics solutions (software systems and services) beyond asset health. This evolution is driving changes in the desired architecture qualities of the toolset; widening the stakeholder pool and influencing priorities; affecting the architecture tradeoffs and decisions; and triggering updates to the product line architecture, the guidance for applying it, and the current prototype of the toolset. We describe our experiences in handling this evolution, assess lessons learned, and discuss potential relevance to other product line scenarios.

Nowadays large volumes of industrial data are being actively generated and collected in various power system applications. Industrial Analytics in the power system field requires more powerful and intelligent machine learning tools, strategies, and environments to properly analyze the historical data and extract predictive knowledge. This paper discusses the situation and limitations of current approaches, analytic models, and tools utilized to conduct predictive machine learning analytics for very large volumes of data where the data processing causes the processor to run out of memory. Two industrial analytics cases in the power systems field are presented. Our results indicated the feasibility of forecasting substations fault events and power load using machine learning algorithm written in MapReduce paradigm or machine learning tools specific for Big Data.

Software Engineering is a complex topic because it encompasses multiple disciplines such as team work, effectively managing change in organizations, understanding technology, understanding software development and its lifecycle, understanding the role of change agents, and also having a good business sense. A method derived from the collaboration between North Carolina State University and ABB, brings diverse techniques that simulate an industrial environment for teaching a senior level Software Engineering course. Eleven elements that have been incorporated to enrich the Software Engineering graduate course are described. The paper also shows how the progressive incorporation of the elements have resulted increased student satisfaction.

This paper presents an approach to analyze the commitment of an organization to a software process improvement (SPI) initiative and to facilitate the interaction among the Sponsor and the Change Agent to improve the probability of its success. Two models are discussed in this paper and they include an organizational readiness evaluation framework (EASE) for SPI and a Game Theory model that facilitates the interaction among players in an SPI initiative. Both models are utilized to improve the organizational readiness and commitment to SPI. Details of both models and a real-world implementation are discussed in detail.

Requirements evolve over the development lifecycle of a software project. Agile practices are designed specifically to address this challenge while showing early and continuous progress towards project goals. Applying an agile approach allows stakeholders to adapt the scope and capabilities of a development release to changing market needs. More recently, an agile approach has been recommended for developing the architecture of software systems, enabling the design to support current requirements and early releases while evolving to meet future expectations. Our experience defining emergent software systems to build a product line architecture for advanced data analytics demonstrates the benefits that can be gained from prioritizing work activities and delaying architecture decisions. This paper proposes a process and ontology for agile architecture development. Only the necessary aspects for each evolutionary release are designed and prototyped, as determined by expectations of the identified application domain scenarios. Feedback from implementing the scenarios using the architecture extends our understanding of the requirements and provides the backlog for successive design iterations.

As an important element of the Smart Grid, Advanced Metering Infrastructure (AMI) systems have been implemented and deployed throughout the world in the past several years. An AMI system connects millions of end devices (e.g., smart meters and sensors in the residential level) with utility control centers via an efficient two-way communication infrastructure. AMI systems are able to exchange substantial meter data and control information between utilities and end devices in real-time or near real-time. The major challenge our research was to scale ABB's Meter Data Management System (MDMS) to manage data that originates from millions of smart meters. We designed a lightweight architecture capable of collect ever-increasing large amount of meter data from various metering systems, clean, analyze, and aggregate the meter data to support various smart grid applications. To meet critical high performance requirements, various concurrency processing techniques were implemented and integrated in our prototype. Our experiments showed that on average the implemented data file parser took about 42 minutes to complete parsing, cleaning, and aggregating 5.184 billion meter reads on a single machine with the hardware configuration of 12-core CPU, 32G RAM, and SSD Hard Drives. The throughput is about 7.38 billion meter reads (206.7GB data) per hour (i.e., 1811TB/year). In addition, well-designed publish/subscribe and communication infrastructures ensure the scalability and flexibility of the system.

People across the world habitually turn to online social media to share their experiences, thoughts, ideas, and opinions as they go about their daily lives. These posts collectively contain a wealth of insights into how masses perceive their surroundings. Therefore, extracting people's perceptions from social media posts can provide valuable information about pertinent issues such as public transportation, emergency conditions, and even reactions to political actions or other activities. This paper proposes a novel approach to extract such perceptions from a corpus of social media posts originating from a given broad geographical region. The approach divides the broad region into a number of sub-regions, and trains language models over social media conversations within these sub-regions. Using Bayesian and geo-smoothing methods, the ensemble of language models can be queried with phrases embodying a perception. Discrete and continuous visualization methods represent the extent to which social media posts within the sub-regions express the query. The capabilities of the perception mining approach are illustrated using transportation-themed scenarios.

Smart cities are powered by the ability to self-monitor and respond to signals and data feeds from heterogeneous physical sensors. These physical sensors, however, are fraught with interoperability and dependability challenges. Moreover, they also cannot shed light on human emotions and factors that impact smart city initiatives. Yet everyday, millions of city dwellers share their observations, thoughts, feelings, and experiences about their city through social media updates. This paper describes how citizens can serve as human sensors in providing supplementary, alternate, and complementary sources of information for smart cities. It presents a methodology, based on a probabilistic language model, to extract the perceptions that may be relevant to smart city initiatives from social media updates. Geo-tagged tweets collected over a two-month period from New York City are used to illustrate the potential of social media powered human sensors.

This paper describes a software estimation technique that can be used in situations where there is no reliable historical data available to develop the initial effort estimate of a software development project. The technique described incorporates a set of key estimation principles and three estimation methods that are utilized in tandem to deliver the estimation results needed to have a robust initial estimation. An important contribution of this paper is bringing together into ONe Software Estimation Tool-kit (ONSET) multiple concepts, principles, and methods in the software estimation field, which are typically discussed separately in the estimation literature and can be employed when an organization does not have reliable historical data. The paper shows how these principles and methods are applied to derive estimates without the need of using complex or expensive tools. A case study is presented using ONSET which was carried out as an estimation pilot study conducted in one of the software development Business Units of ABB. The results of this pilot project provided insights on how to implement ONSET across ABB software development business units. Practical guidance is offered in this paper on how an organization that does not have reliable historical data can begin to collect data to use in future projects using ONSET. In contrast to many papers that describe estimation approaches, this paper explains how to use a combination of judgment-based and model-based methods such as the Planning Poker, Modified Wideband Delphi, and Monte Carlo simulation to derive the initial estimates. Once an organization begins collecting reliable historical data, ONSET will provide even more accurate estimation results and a smoother transition to the use of model-based estimation methods and tools can be achieved.

ABB is a multi-national corporation that is developing a new generation of products based on the concept of Industrial/sup IT/. This concept provides a common integration platform for product interoperability. As Industrial/sup IT/ enabled products are developed across ABB, software reuse must be considered. Component based software development (CBSD) is an effective means to improve productivity and quality by developing reusable components. Measuring the economic benefits and performing sensitivity analyses of CBSD scenarios in the development of Industrial/sup IT/ products is important to improve efficiency. This paper presents a model that allows project leaders to evaluate a variety of software development scenarios. The model is based on a goal-question-metrics (GQM) approach and was developed at the ABB corporate research laboratories.

Current software environments face increased pressure to develop products under evolving requirements, changing technologies, scarce human resources, and the need to develop high quality applications. ABB has similar pressures, especially as the organization embraces the Industrial IT initiative that aims at the development of interoperable and intelligent products. A new generation of software development lifecycle models has emerged lately called "Agile" and they embrace change, reduce development cycle time, and attempt a useful compromise between no process and too much process. The Agile Development in Evolutionary Prototyping Technique (ADEPT) presented in this paper, was developed at the ABB US Corporate Research Center, and incorporates "Agile" practices to streamline the technology development lifecycle of Industrial IT products.

Evolutionary prototyping focuses on gathering a correct and consistent set of requirements. The process lends particular strength to building quality software by means of the ongoing clarification of existing requirements and the discovery of previously missing or unknown requirements. Traditionally, the iterative reexamination of a systems requirements has not been the panacea that practitioners sought, due to the predisposition for requirements creep and the difficulty in managing it. The paper proposes the combination of evolutionary prototyping and an aggressive risk mitigation strategy. Together, these techniques support successful requirements discovery and clarification, and they guard against the negative effects of requirements creep. We embody these techniques in a comprehensive software development model, which we call the EPRAM (Evolutionary Prototyping with Risk Analysis and Mitigation) model. The model was intentionally designed to comply with the Level 2 Key Process Area of the Software Engineering Institute's Capability Maturity Model. Validation is currently underway on several software development efforts that employ the model to support the rapid development of electronic commerce applications.

With the introduction of the Industrial/sup IT/ vision at ABB, the coordination of hardware and software development is a crucial aspect that the organization needs to address. Software development and hardware production processes need to be coordinated into one coherent lifecycle model to optimize development. This paper presents some basic concepts in software and hardware development lifecycles and paves the road for a more comprehensive approach for a system development lifecycle.

The paper explains the importance of distributed manufacturing networks (DMNs) as a strategy used by manufacturing companies such as Asea Brown Boveri to gain competitive advantage. An important aspect of DMNs is the development of effective tools that allow geographically dispersed teams to design and manufacture products. The importance of distributed concurrent engineering (DCE) is explained. Manufacturability is presented as a salient aspect of a DCE approach. The architecture of a Web-portal-based distributed concurrent engineering system that analyzes the manufacturability of a product is presented.